> ## Documentation Index
> Fetch the complete documentation index at: https://docs.tqx.ai/llms.txt
> Use this file to discover all available pages before exploring further.

# Strategy code

> Understand the Qube strategy runtime and market-specific coding rules.

Qube runs strategy code through a backtest engine. Keep the code focused on trading logic. Set the run period, capital, frequency, and other simulation controls in the canvas or chat tools.

## Runtime lifecycle

Use this lifecycle for a strategy:

```python theme={null}
def initialize(context):
    # Set accounts, symbols, parameters, and empty caches.
    init_market_data(context)


def init_market_data(context):
    # Load the required history once and prepare reusable data.
    pass


def handle_data(context, data):
    # Read prepared data, create signals, and submit orders.
    pass
```

The exact imports and order functions depend on the market. Use the market-specific generation flow when you create code with Qube.

## Rules that apply to every market

* Load full history in `init_market_data`, not on every bar.
* Initialize custom `context` fields before reading them.
* Keep `handle_data` focused on signal checks and orders.
* Do not use future data or future-return fields to create a signal.
* Do not call chat tools such as `set_backtest_params` or `run_backtest` from strategy source.
* Check for missing data and return explicitly instead of hiding errors.

## Market-specific differences

| Market    | Data and execution focus                                                |
| --------- | ----------------------------------------------------------------------- |
| Hong Kong | Use HK data APIs, `.HK` symbols, the HK calendar, and valid board lots. |
| US        | Use US data APIs, `.NB` symbols, and the US calendar.                   |

<Info>
  For minute strategies, the backtest frequency is `1M`. The history API may use a different spelling such as `1m`; do not confuse the task frequency with the data query period.
</Info>

## Keep simulation settings outside the source

The canvas or chat controls the initial capital, dates, frequency, commission multiplier, slippage, and benchmark. A strategy should read the runtime context and implement the trading rules instead of hard-coding a user's one-off backtest request.